A two-stage stochastic programming approach for generation and transmission maintenance scheduling with risk management

In this article, we study the generation and transmission maintenance scheduling problem under uncertainty. We propose a two-stage optimization model with the first stage for weekly maintenance scheduling and the second stage for hourly economic power dispatch. To address the future uncertainties associated with renewable energy penetration and electricity demand, we formulate the problem as a two-stage stochastic mixed-integer programming model and incorporate Conditional Value at Risk (CVaR) to control the risk of having extreme loss of demand. To facilitate practical implementation, we apply the Benders decomposition algorithm tailored for parallel computing as the solution approach for the problem. The maintenance decisions, computational performance, and optimality of the model are evaluated by case studies on IEEE test instances. An extensive sensitivity analysis of CVaR related parameters is performed to illustrate their impact on decisions and risk management. The results show that the proposed risk-constrained model can provide effective annual maintenance plans for both generators and transmission lines on a weekly basis, and the Benders decomposition algorithm is able to solve large-scale problem instances efficiently.